Unverified50% confidenceFactExact time
Developer datavorous shared an experiment on Reddit: by fine-tuning golfed JavaScript code and self-scoring 40,000 samples for RLHF, training a model to explore the syntax of the Demoscene and the possibilities of unfinished code.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/1/2026
First Seen
Valid until: 1/30/2027
Sources
Related Entities
Related Claims
UnverifiedA developer shared on Reddit their journey of learning reinforcement learning (RL) from scratch and decided to organize the learning process into a series of short notes published on their personal website.70% similarUnverifiedAn aperiodic-order researcher (jarekduda) posted a mini dataset containing 9 problems on Reddit to explore reinforcement learning with verifiable rewards (RLVR).69% similarUnverifiedThe test comes from an experiment published by an individual Reddit developer and has not yet been independently replicated by a third party; the author has open-sourced the complete experimental configuration, request contents, and confusion matrix on GitHub Gist.69% similarUnverifiedThis article is based on experimental results shared by a Reddit user, and the specific implementation details and code were not provided in the original text.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1122305API
curl https://kongchang.com/api/v1/knowledge/claims/1122305MCP
get_claim(id=1122305)